Andrej Karpathy

Because if you want to improve a system by 10%, it costs some amount of work.

And if you want to 10x improve the system, it doesn't cost...

you know, a hundred X amount of the work.

And it's because you fundamentally change the approach.

And if you start with that constraint, then some approaches are obviously dumb and not going to work.

And it forces you to reevaluate.

And I think it's a very interesting way of approaching problem solving.

Yeah.

I mean, I think a good example here is, you know, the deep learning revolution in some sense, because you could be in computer vision at that time during the deep learning sort of revolution of 2012 and so on.

You could be improving a computer vision stack by 10%, or we can just be saying, actually, all of this is useless.

And how do I do 10x better computer vision?

Well, it's not probably by tuning a hog feature detector.

I need a different approach.

I need something that is scalable.

Going back to Richard Sutton's understanding of the philosophy of the bitter lesson, and then being like, actually, I need a much more scalable system, like a neural network, that in principle works.

And then having some deep believers that can actually execute on that mission and make it work.

So that's the 10x solution.

I think the tough thing with timelines of self-driving, obviously, is that no one has created self-driving.

So it's not like, what do you think is the timeline to build this bridge?

Well, we've built a million bridges before.